I think that's one of the biggest roadblocks and victories on their part that is going to be, you know, incredibly hard to root out. How do we communicate that these things are clearly statistical, mathematical constructions and phenomenon that are not in any way, shape or form intelligent? And that the gulf between these narrow applications of these models,. which may have some practical value to something that is actually intelligent is such a huge gulf. We haven't also been having larger, deeper thoughts about how we use our language," he says.